The procurement of goods and services in Public Administration is crucial for achieving institutional goals, with a focus on financial responsibility and trans- parent decision-making. In Costa Rica, public procurement is centralized through the Integrated Public Procurement System (SICOP). This study concentrates on goods procurement, aiming to identify successful contracts and detect anomalies. Machine Learning techniques, particularly under unsupervised approaches, enhance anomaly detection. The Principles of Integrity in Public Procurement Procedures from the Organisation for Economic Co-operation and Development (OECD) guide the evaluation process, emphasizing good procurement management, prevention of misconduct, and transparency. Various indicators, such as realistic budget estimation and objection rates, are utilized. Rapid procurement processes and price alterations may signal vulnerabilities and misconduct, highlighting the need for transparency and market awareness. Discovering its patterns is critical for accurate results, as different models respond differently to datasets and sample size changes. Emphasis should be placed on similar population clusters to avoid detecting natural anomalies. Implementing management mechanisms and employing data cleaning techniques are recommended to address data management errors.